4 ms·
2 years of RLHF I don’t see reasoning. It’s clearly a text-similarity game, plus training of course. My mental model to work best is not that I’m interacting wi
by meroes 1y ago
2 years of RLHF I don’t see reasoning. It’s clearly a text-similarity game, plus training of course. My mental model to work best is not that I’m interacting with a thing with reasoning. That differs from my mental model of interacting with humans. I would be worse at my job if I interacted with LLMs as if they reasoned.
- ninetyninenine 1y agoWhere's the evidence? You have none. Again you made a leap here with ZERO evidence which is equivalent to an hallucination.